Local Weighted Directional Pattern for Facial Expression Recognition

  • 王 國明

Student thesis: Master's Thesis

Abstract

In this thesis a weighted local directional texture descriptor is proposed which is applied in facial expression recognition We partition the face images into several regions calculate the distribution of the structure of micro patterns using the directional information for each region to represent the facial texture feature with the concept of weight and the horizontal projection of the result of edge detection is also used to help identify facial expression Finally SVM (support vector machine) is used to classify the facial expression We use three databases to test the performance of the proposed facial expression recognition system There are seven expressions in both JAFFE database and Cohn-Kanade database which are neutral happiness sadness surprise fear anger and disgust For CMU-PIE database there are two expressions which are neutral and smiling As the experimental result the proposed method outperforms the other methods in facial expression recognition
Date of Award2014 Jul 25
Original languageEnglish
SupervisorShen-Chuan Tai (Supervisor)

Cite this

'